Lists Home |
Date Index |
From: Dave Pawson [mailto:firstname.lastname@example.org]
At 14:19 26/06/2003 -0500, Bullard, Claude L (Len) wrote:
>>So here is a fun puzzle for the cognoscenti. Can John Sowa's
>>rules be applied to XML using XSLT or am I just taking too few
>>or too many medications? ;-)
>>Or does this require RDF?
>Not claiming to be anywhere near coginscenti, but hey Len,
>What the heck can RDF do that is even close to XSLT?
Nothing I think. What happens is that the means
for establishing the patterns has to be notated
somewhere. XSLT is the transform engine, but
it also stores the output patterns. There is an XML
version of RDF of course.
I haven't worked this all out. I am still studying
Sowa's work, but he encourages this line of thinking
as long as I understand that FOL is not an 'exotic
technology'. To me the fascination is with analogical
reasoning systems and how they could be implemented with
basic XML technologies. It could be a poor man's AI and
it can be applied to domains we are looking at in HumanML.
I asked Sowa about the use of XSLT for this. John said:
"That sounds promising.
Just one comment:
> ... can in part
> be achieved with non-exotic technologies
> such as XSLT (the XML transformation language)
> and smartly designed XML. Note that adding
> logic to that is not difficult or uncommon.
I would avoid the words "exotic" and "non-exotic".
First-order logic is the simplest language you can
imagine -- it is a subset of every natural language,
and you only need a half-dozen operators, namely,
and, or, not, if-then, every, and some. (You could
actually get by with only two operators, but then
it becomes much harder to read, write, and use.)
I agree that some notations for FOL are more painful
to read then others. But whenever people try to invent
so called "non-exotic" notations, they end up with
horrible notations such as SQL and OCL, which are
vastly more cumbersome, inflexible, and unreadable
than the so called exotic notations.
If people want something they can read, I recommend
controlled natural languages, of which ACE is one:
I also discuss these issues in the ontometa article;
And for a summary of the textbook notations, see